Surgical risk prediction tools are a useful tool in the stratification of clinical management of patients as well as providing a benchmark for adjusting the outcomes reflected in a registry or in the experience of a center. Typically, the EuroSCORE II and STS-score have been the reference systems, but the representation of different types of pathology or population subgroups in them has limited their validity in other contexts. Thus, the representation of women with coronary pathology, the treatment of isolated tricuspid insufficiency, patients with chronic liver disease, among others, did not demonstrate adequate representation in these databases, having derived initiatives such as the TRI-SCORE or the application of the MELD score, which had already been previously analyzed in the blog, as better predictors of surgical risk and outcomes.
It is worth noting that these types of risk scales are indicative, distribute patients in a graduation of low, moderate, or high risk, but the value they offer should not be considered to the letter for each particular case, where a risk range should be taken into account when informing the patient, once estimated based on their clinical condition and particular characteristics of each case. And what we consider low, intermediate, or high also varies from one pathology to another, with different degrees of definition. We know that for revascularization surgery a risk of 4% would be high, while for aortic valve replacement, as is widely known, it would be the limit to start considering a moderate risk.
Type A dissection also constitutes a special context where EuroSCORE II shows notable gaps. It is an entity that involves systematically marking emergency items and aortic surgery, two of the ones that have the greatest impact on surgical mortality. However, the weight that has been given to them in EuroSCORE II comes from the aggregation of cases included in its development in which, the presence of these variables, was not necessarily due to cases of type A aortic dissection. In addition, the interaction that these variables present in the model, with each other and with others such as age, renal dysfunction, need for multiple surgical procedures, or critical perioperative situation, probably is not specific to the context of type A aortic dissection but derived from the entire spectrum of cardiovascular surgery. Finally, there are peculiarities such as the location of the entry door, the extension of the dissection, or the presence of territories with poor perfusion that are important determinants of the patient’s prognosis and that are not considered.
Czerny et al., based on the powerful database of the German registry GERAADA, developed a predictive model that today, the authors of the work try to validate in the experience of their institution, in a context with similarities and differences with the original, such as the American system (Pennsylvania). Between 2010 and 2021, they included the retrospective experience with 689 patients operated on for type A aortic dissection. The overall mortality of the series was 12% at 30 days, involving 80% type I dissections, with 27% of patients hemodynamically unstable, with poor perfusion of at least one territory in 41% and with significant aortic insufficiency in 23%.
The GERAADA scoring system comprises a stratification into low risk <15%, intermediate 15-30%, or high risk >30%. It is assumed that the skill of the Pennsylvania group is enviable, since it is impossible to believe that the average mortality is entrenched in a profile of low risk due to the characteristics of the operated patients. Although the American system is more selective than the European environment in the choice of candidates for surgery, it is more likely that the data is due to the good results that are well below the predicted risk. This is proof of the low utility of these risk assessment systems in predicting specific risks (adjustment), which has nothing to do with the true function, the stratification of patients.
And in this aspect is where he demonstrated good discrimination capacity, with an area under the ROC curve of 0.76. The authors conducted subanalyses determining the discrimination capacity in different risk subgroups within type A aortic dissection. Thus, the GERAADA score demonstrated the best discrimination for the presence of the primary entry in the aortic arch (area under the ROC curve of 0.86) and for the presence of significant aortic insufficiency (area under the ROC curve of 0.82). The worst discrimination occurred in cases of reoperation due to the presence of previous surgery (area under the ROC curve of 0.69) and the need for resuscitation prior to surgery (area under the ROC curve of 0.67). As for the age groups, the best discrimination occurred for patients between 50 and 59 years old (area under the ROC curve of 0.81), being poorer in patients with extreme ages >80 years old (area under the ROC curve of 0.64).
The authors conclude that the GERAADA scoring system is a practical and easily accessible tool to reliably estimate the 30-day mortality risk of patients undergoing surgery for acute type A aortic dissection.
COMMENTARY:
The existence of surgical risk estimation scales is, for the generations that we currently find ourselves in active, inherent to the surgery we perform. Probably, derived from the high standardization of most procedures, it is possible to apply this type of methodologies that, in other fields of surgery, would be unthinkable. Also, in the DNA of those generations is the work with databases and records, which is a desire to compare ourselves with others and, now more than ever, with the interventional competitor.
The GERAADA score (https://web.imbi.uni-heidelberg.de/geraada-score/) is a logistic model analogous to EuroSCORE II that contemplates the following variables: age, sex, need for resuscitation prior to surgery, previous cardiac surgery, intubation and mechanical ventilation upon patient reception, preoperative catecholaminergic support, presence of aortic insufficiency, presence of poor perfusion (with the same criteria as the TEM classification), presence of preoperative neurological deficit (hemiparesis/plegia), extension of type A dissection and location of the entry door (again compatible with the TEM classification).
Naturally, models like that of GERAADA have been developed in environments and with the experience of centers that, probably, exceed those of many others, including our country. However, it seems realistic since it may overestimate the risk in patients undergoing surgery in experienced centers in aortic pathology with results well below the predicted risk. And, I repeat, in calibration the important thing is not the prediction but the stratification when making decisions with individual patients. For this reason, a context of similar gravity as type A aortic dissection and surgery of such complexity as that required, involves too many details that can “make you go from joy to tears” in a matter of seconds, and mark the prognosis of the patient. In this way, it is necessary to incorporate them, as far as possible, into these models in the form of those clinical conditions in which such complications or adverse events can occur more frequently. In fact, it can be seen how the predictive capacity of the score changes in the subanalyses when considering variables that condition the technical complexity or the variability of the clinical context such as the presence of an entry door in the aortic arch, the age of the patient or hemodynamic instability.
German tools remain reliable. Czerny and the GERAADA registry provide us with a useful tool in decision-making in a context as adverse as aortic dissection. The authors of this work have validated it in the experience of a high-level American center. It would be very good to replicate similar experiences in our environment, for example, combining it with the data of the Spanish Registry of Cardiac Surgery (RECC)… who dares?
REFERENCE:
Berezowski M, Kalva S, Bavaria JE, Zhao Y, Patrick WL, Kelly JJ, et al. Validation of the GERAADA score to predict 30-day mortality in acute type A aortic dissection in a single high-volume aortic centre. Eur J Cardiothorac Surg. 2024 Feb 1;65(2):ezad412. doi: 10.1093/ejcts/ezad412.
